On Fri, Nov 3, 2017 at 9:05 AM, Matthias Leopold <matthias.leopold@xxxxxxxxxxxxxxxx> wrote:
Am 2017-11-03 um 02:44 schrieb Jason Dillaman:
On Thu, Nov 2, 2017 at 11:34 AM, Matthias Leopold <matthias.leopold@meduniwien.ac.at <mailto:matthias.leopold@meduniwien.ac.at >> wrote:
Hi,
i'm trying to set up iSCSI gateways for a Ceph luminous cluster
using these instructions:
http://docs.ceph.com/docs/master/rbd/iscsi-target-cli/
<http://docs.ceph.com/docs/master/rbd/iscsi-target-cli/ >
When arriving at step "Configuring: Adding a RADOS Block Device
(RBD)" things start to get messy: there is no "disks" entry in my
target path, so i can't "cd /iscsi-target/iqn.2003-01.com
<http://iqn.2003-01.com>.redhat.iscsi-gw:<target_name>/disks /". When
i try to create a disk in the top level "/disks" path ("/disks>
create pool=ovirt-default image=itest04 size=50g") gwcli crashes
with "ValueError: No JSON object could be decoded" (there is more
output when using debug but i don't think it matters). More
interesting is /var/log/tcmu-runner.log, it says consistently
[DEBUG] handle_netlink:207: cmd 1. Got header version 2. Supported 2.
[DEBUG] dev_added:768 rbd/ovirt-default.itest04: Got block_size 512,
size in bytes 53687091200
[DEBUG] tcmu_rbd_open:581 rbd/ovirt-default.itest04: tcmu_rbd_open
config rbd/ovirt-default/itest04/osd_op_timeout=30 block size 512
num lbas 104857600.
[DEBUG] timer_check_and_set_def:234 rbd/ovirt-default.itest04: The
cluster's default osd op timeout(30.000000), osd heartbeat grace(20)
interval(6)
[DEBUG] timer_check_and_set_def:242 rbd/ovirt-default.itest04: The
osd op timeout will remain the default value: 30.000000
[ERROR] tcmu_rbd_image_open:318 rbd/ovirt-default.itest04: Could not
open image itest04/osd_op_timeout=30. (Err -2)
The error is that ceph-iscsi-config has instructed tcmu-runner that the name of the image is "itest04/osd_op_timeout=30". We changed the delimiter for separating optionals from "/" to ";" and that is what your version of tcmu-runner is expecting. Upgrade to the latest available version of ceph-iscsi-config from here [1].
thank you very much, now this problem is gone and i happily run into the "missing kernel attribute: qfull_time_out" issue ;-)
i read your comments in https://bugzilla.redhat.com/show_bug.cgi?id=1503711 and the http://lists.ceph.com/pipermail/ceph-users-ceph.com/2017- thread, so i guess i'll have to wait for newer upstream and subsequently RHEL/CentOS kernels. i can only say we would be very happy if these patches could already be included in the 7.4.z kernels.November/022042.html
This is a 4.14-rc7 kernel w/ Mike's patches applied [1] if you wanted to play.
one minor issue i noticed is that gwcli doesn't like hyphens in pool names. i don't know if this check is really necessary, we just happen to have hyphens in our pool names, so we would have to change that
That should hopefully be fixed by this PR [2] once it's merged (haven't had a chance to test it yet).
thanks a lot for your good work
matthias
[2] https://github.com/ceph/ceph-iscsi-cli/pull/41
Jason
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
